Simultaneous EEG/fMRI Reveals Spatiotemporal Correlates of Baseball Pitch Recognition

Abstract

pitch trajectories (Fastball, Curveball, and Slider), while we simultaneously recorded BOLD fMRI at 3x3x3 mm, using a 3T scanner, and scalp EEG, using our custom-built 43channel EEG system [6]. Each trial consisted of a letter cue (“F”-Fastball, “C”-Curveball, “S”-Slider) and countdown bar, the simulated pitch, and then a (+/-) feedback sign to indicate if the subject was correct on the trial (Fig. 1) . Subjects were instructed to respond when the pitch matched the cue and before the pitch ended. There were a total 450 trials over 5 scans with a 60/40 split between Go/NoGo. We performed standard preprocessing on all data [6]. We used logistic regression with a sliding window [7] on stimulus-locked epochs to discriminate between combinations of Correct Go,Correct NoGo, and Incorrect NoGo trials using the EEG data. We first analyzed the BOLD signal using a traditional event related analysis. For the single-trial variability analysis, we analyzed the BOLD signal by correlating the trial-to-trial variability of the EEG classifiers within each class with the BOLD data, similar to the approach of Goldman et al. (see [6] for details). Results are presented at the group level.